def __init__(self):
fileParser = FileParser()
self.__table = fileParser.parse("graph")
self.__maxSize = fileParser.getMax()
self.__price = []
self.__date = []
self.__range = 365
def predict():
fileParser = FileParser()
table = fileParser.parse("predictor")
maxSize = fileParser.getMax()
#print(maxSize)
epsilon = .6 #percent error
dataContainer = []
for i in range(0, maxSize):
openValue = table.iloc[i, 0]
closeValue = table.iloc[i, 1]
percentDifference = (closeValue - openValue) / openValue * 100
dataContainer.append(
percentDifference
) #stores the percent difference between the closed and open value
if DEBUG:
print(table)
accumMatch = 0
accumIncrease = 0
#averages the current 3 percent differences
currentAvg = (dataContainer[maxSize - 1] + dataContainer[maxSize - 2] +
dataContainer[maxSize - 3]) / 3
if DEBUG:
print("\"\"\"DEBUG START\"\"\"")
print("LAST THREE DIFFERENCE:")
#print("1.", dataContainer[0])
print("1.", dataContainer[maxSize - 1])
print("2.", dataContainer[maxSize - 2])
print("3.", dataContainer[maxSize - 3])
print("CURRENT AVG:", currentAvg)
for i in range(2, maxSize - 1):
#gets the averages of all the past 3 percent differences and checks if the trend
#is similar to the current average with margin of error (epsilon)
pastAvg = (dataContainer[i] + dataContainer[i - 1] +
dataContainer[i - 2]) / 3
if currentAvg - epsilon < pastAvg and currentAvg + epsilon > pastAvg:
accumMatch += 1
#within those similar trends, check how much of those events actually increased afterwards
if dataContainer[i + 1] > 0:
accumIncrease += 1
#This means that there was no similar trends, then give a 50:50 prediction
if accumMatch == 0:
return (.5, .5)
probabilityIncrease = accumIncrease / accumMatch
probabilityDecrease = (accumMatch - accumIncrease) / accumMatch
if DEBUG:
print("ACCUM INCREASE:", accumIncrease)
print("ACCUM MATCH:", accumMatch)
print("PROBABILITY TO INCREASE:", probabilityIncrease)
print("PROBABILITY TO DECREASE:", probabilityDecrease)
print("\"\"\"DEBUG END\"\"\"")
return (probabilityIncrease, probabilityDecrease)
